Using singular value decomposition

[JoFAM]

Try to reformulate everything in terms of SVD as opposed to eigen decomposition in an attempt to avoid complex results for borderline cases.

The code in in exampleCode/Benchmarking.R shows the improvement of using crossprod and tcrossprod, but the result consists of complex numbers. This shouldn't be happening.

[JoFAM]

Been looking up a few things, but there seems no general way of translating one into the other for my poor mathematical brain. Just adding this link for reference, as they look at it in terms of gene expression data

http://web.mit.edu/be.400/www/SVD/Singular_Value_Decomposition.htm

@MichielStock pointers or help appreciated.